BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Detecting Sybils without Graphs - Ben Zhao\, University of Califor
 nia
DTSTART:20130206T110000Z
DTEND:20130206T120000Z
UID:TALK43225@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:Fake user accounts are a growing problem today for numerous on
 line social networks.  For years\, researchers have relied on community de
 tection algorithms to propose algorithms and systems that detect these Syb
 il accounts.  However\, recent measurement work showed that attackers inte
 ntionally avoid forming communities\, calling the efficacy of these system
 s into question.\n\nIn this talk\, I will present results of two projects 
 focused on using novel methods to detect fake Sybil accounts without relyi
 ng on social graph structures.  First\, I will talk about our work explori
 ng the use of crowdsourcing as a core component in a scalable Sybil detect
 ion system.   We carry out a large user study analyzing the ability of cro
 wdsourcing workers to quickly and cheaply detect fake account profiles\, u
 sing a large corpus of ground-truth Sybil accounts from the Facebook and R
 enren networks.  We analyze detection accuracy by both "experts" and "turk
 ers" under a variety of conditions\, and find that while turkers vary sign
 ificantly in their effectiveness\, experts consistently produce near-optim
 al results.  We use these results to drive the design of a multi-tier crow
 dsourcing Sybil detection system.  Using our user study data\, we show tha
 t this system is scalable\, and can be highly effective either as a standa
 lone system or as a complementary technique to current tools.  Second\, I 
 will present early results of a new study on Sybil detection using models 
 of user clickstream events.  We show that legitimate and Sybil users diffe
 r dramatically in user-generated events\, and propose a unsupervised learn
 ing system for effectively identifying Sybil users based on user actions.
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR